Dissociative and molecular oxygen chemisorption channels on reduced rutile TiO2(110): An STM and TPD study
Introduction
Titanium dioxide (TiO2) is a reducible transition metal oxide that is widely used in a number of technological fields such as photocatalysis, heterogeneous catalysis, solar cells, hydrophilic films, gas sensors, waste remediation, and biocompatible materials [1], [2], [3], [4], [5], [6], [7], [8]. Among these applications particularly in areas such as photocatalysis, photodegradation of organic pollutants and the photo-generation of hydrophilic films, the interaction of O2 with TiO2 plays an important role [1], [3], [5], [7], [8]. For example, O2 is a common oxidant and is also used in photocatalysis as a scavenger of the photo-excited electrons to prevent negative charge accumulation on the surface of the catalysts [1], [2], [3]. Furthermore, the interaction of O2 with TiO2 is interesting with a view to the possible photodynamic therapy of cancer, where the O2 molecules play a vital role as oxidizing species [1]. To enhance the efficiencies of the applications listed above, it is essential to improve our understanding of how O2 interacts with TiO2 surfaces. The promising applications of TiO2-based materials have spurred tremendous research both in fundamental as well as in more applied fields.
In early surface science studies under ultra-high vacuum (UHV) conditions, the interaction of molecular O2 with TiO2 single crystals was studied most frequently using photon-stimulated desorption (PSD) and temperature-programmed desorption (TPD) [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22]. In all these works the rutile TiO2(110)–(1 × 1) surface (cf. Fig. 1), which is the most stable surface of rutile, and which is often considered the model system for transition-metal oxide surfaces [4], [6], [17], [23], [24], [25], [26], [27], was studied. Very valuable information has been gained by using these desorption techniques on the O2 interaction with rutile TiO2(110)–(1 × 1). For example, O2-PSD studies by the Yates group [8], [9], [10], [16], [17] have revealed that the O2-PSD occurs in UHV on O2-exposed TiO2(110) very much as it is observed on high-surface-area TiO2 powder materials [28], [29], [30]. In addition, the thermal chemistry has been examined by Henderson and co-workers by combining detailed TPD studies with electron energy loss spectroscopy (EELS) measurements [12], [13], [14]. These studies disclosed the complex nature of the O2–TiO2(110) interaction and showed that dissociative and molecular channels co-exist on reduced TiO2(110) crystals. The desorption of O2 at ~ 410 K was observed for TiO2(110)–(1 × 1) samples characterized by ~ 8%ML (monolayer) bridging oxygen (Obr) vacancies after O2 adsorption at temperatures lower than ~ 180 K [12], [13], [14], [16]. Interestingly, no sign of scrambling was found for the O2 molecules that desorb in the TPD peak centered at ~ 410 K neither with the Obr atoms on the surface nor with other O2 molecules [12], [15]. EELS measurements indicated that charge is transferred from the TiO2(110) surface to O2 molecules adsorbed at 120 K and suggested the stabilization of a superoxo species (singly charged O2) on the surface [12], [13], [14]. Recent low-temperature TPD studies by Dohnálek et al. [18] and Kimmel and Petrik [20] showed that physisorbed O2 desorbs from the TiO2(110) surface at temperatures below 100 K. Utilizing the desorption of physisorbed O2, it is possible to distinguish between physisorbed and chemisorbed forms of O2, and the amount of chemisorbed O2 can be quantified using liquid He cooled samples [20].
However, based alone on desorption and other spectroscopic techniques insights into reactions on the surface can only be deduced indirectly, which may explain why some of the previous models of the O2–TiO2(110) interaction are rather speculative. For example, Henderson et al. suggested that O2 adsorption onto reduced TiO2(110)–(1 × 1) at ~ 120 K would lead to the stabilization of up to three O2 molecules per Obr vacancy [12], and Kimmel and Petrik reported that each Obr vacancy would stabilize two O2 molecules at low temperatures [20]. On the other hand, STM studies by Wendt et al. reported that the O2 molecules dissociate at Obr vacancy sites [Fig. 2(a)] even at temperatures as low as ~ 120 K [26], [31], [32], at variance with the models previously proposed [12], [20]. By means of TPD, it is rather difficult to gather information about O2 dissociation on the reduced TiO2(110) surface. Yet, the existence of O adatoms (Oot) on the five-fold-coordinated Ti (5f-Ti) sites (cf. Fig. 1, Fig. 2), created by O2 dissociation, has been inferred from TPD studies addressing the chemistry of water and ammonia on O2-exposed TiO2(110) single crystals [11]. The TPD peaks ascribed to the first monolayer (ML) of water and ammonia, respectively, were found to shift to higher temperatures on O2-predosed TiO2(110) crystals and explained through additional H bonds to the Oot adatoms on the surface [11]. Using a desorption technique, more direct evidence of the Oot adatoms was found only recently by the Yates group [21]; Lee et al. used electrons with 210 eV to stimulate the desorption of Oot adatoms and reported that these Oot adatoms on the TiO2(110) have a very high ionic cross section for O+ production [21].
The technique of choice to study the dissociative adsorption of O2 on reduced TiO2(110) is STM. Previous STM studies have unraveled a surprisingly rich and complex interaction of O2 with the reduced TiO2(110) surface [4], [6], [25], [26], [27], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42]. For example, Li et al. reported the formation of irregular networks of pseudo-hexagonal rosettes and [001]-oriented strands on originally flat TiO2(110)–(1 × 1) surfaces through the oxidation of vacuum-annealed crystals at 470 K–830 K [36]. In addition, Du et al. [39] showed that O2 dissociation at Obr vacancy sites (the first O2 dissociation channel) at room temperature (RT) leads to Oot adatoms left behind in the Ti troughs that are separated by one lattice constant along the [001] direction away from the original Obr vacancy positions [Fig. 2(b)]. Moreover, results obtained by our group provided evidence of a non-vacancy-assisted O2 dissociation channel that occurs in addition to the first, vacancy-assisted dissociation channel [32]. Upon O2 exposure at RT, this second O2 dissociation channel leads to the formation of paired Oot adatoms on next-nearest 5f-Ti sites in the Ti troughs [Fig. 2(c)]. Recently, Du et al. confirmed the formation of Oot adatom pairs on the surface of reduced TiO2(110) crystals upon O2 exposure at RT [42]. The occurrence of the second O2 dissociation channel is a strong indication of the existence of charge donors in the near-surface region, since without charge donation the adsorption and dissociation of O2 on TiO2(110) is predicted to be energetically unfavorable [31], [32], [43], [44], [45].
The exact nature of the charge donors on the surface of reduced TiO2(110) crystals is still an issue of ongoing research [33], [44], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56]. For a long time the prevailing opinion in the literature was that the chemistry on reduced TiO2(110) is governed by Obr vacancies [4], [6], [8], [23], [24], [57], [58], [59], [60], [61]. Consequently, it was thought that the Ti3d defect state in the band gap observed on reduced TiO2 samples [4], [12], [24], [48], [59], [60], [62], [63], [64] was caused by missing surface O atoms. However, this belief was recently disputed [32], because a number of experimental studies provided solid evidence of the existence of Ti interstitials in the bulk of reduced TiO2(110) samples [32], [34], [36], [37], [65], [66], [67], which could also lead to the defect state in the band gap [50], [51], [68], [69], [70]. In a recent paper, we reported on valence band spectroscopic measurements showing that the Ti3d defect state remains almost unchanged on nearly defect-free TiO2(110) surfaces compared to surfaces with defects, Obr vacancies and H adatoms (or OHbr groups), respectively. Based on this and additional results we proposed that the Ti3d defect state arises predominantly due to the extra charge originating from Ti interstitials in the near-surface region rather than from Obr vacancies [32]. On the contrary, two recent studies reported that electron-donating surface defects, Obr vacancies and H adatoms, rather than Ti interstitials control the reactivity of TiO2(110) [53], [55]. However, in both these studies TiO2(110) crystals were irradiated with electrons in order to create additional surface defects. Because electron irradiation does not exclusively create Obr vacancies, but also other, unidentified defects in the near-surface region of TiO2(110) crystals [56] the conclusions of these two studies are debatable. Further research is needed to clearly disentangle surface and near-surface contributions to the defect state and to improve our understanding of the redox chemistry on the surfaces of bulk-reduced TiO2 crystals.
In the present article we report on systematic and detailed STM studies on the prevailing O2 dissociation channel on reduced TiO2(110)–(1 × 1) as a function of the adsorption temperature. The STM studies reveal that the second O2 dissociation channel is an activated process requiring temperatures higher than 150 K–180 K and that the O2 dissociation reaction on reduced TiO2(110) is self-limiting. The STM results are supported by TPD data that allow us to directly compare the here presented new results with previously published spectroscopic data. The results are explained by invoking a competition between dissociative and molecular O2 interaction channels for the electronic charge in the near-surface region, and a new model for the interaction of molecular O2 with reduced TiO2(110) crystals is proposed. From the STM and TPD results we infer that charge donation from bulk defects to adsorbed oxygen species plays a crucial role, however, the kinetics of the two O2 dissociation channels as well as the kinetics of the diffusion and reaction of Ti interstitials must be considered as well.
Section snippets
Experimental
The STM and TPD experiments were carried out in a UHV chamber with a base pressure in the low 10−11 Torr range equipped with a homebuilt, variable-temperature Aarhus STM, a quadrupole mass spectrometer (QMS), and standard facilities for sample preparation and characterization [71], [72]. The Aarhus STM used for this study is capable of fast scanning and high resolution in a temperature range between 100 and 400 K. Electrochemically etched tungsten tips were used in all the STM experiments. The
Combined STM and TPD data
Fig. 3 shows empty state STM images acquired from the same TiO2(110) crystal (sample #1) at different reduction states, low- and medium-reduced, respectively. As the crystal was low-reduced [Fig. 3(a)], steps were running preferentially parallel to the [1–11] and [1–1–1] directions, while steps in the [001] direction occurred less frequently. Steps running parallel to the [1–11] and [1–1–1] directions are uniform and smooth, whereas steps running parallel to the [001] direction are often rough,
TPD results
The O2-TPD data presented in Fig. 4, Fig. 6 are in good agreement with previously published data showing O2 desorption peaking at ~ 410 K from sputtered and vacuum-annealed TiO2(110) crystals [12], [14], [16], [20]. As in the previous studies we found the O2 desorption temperature to be invariant with O2 coverage to within ± 5 K, suggesting first-order desorption kinetics rather than second-order O atom recombination kinetics. Furthermore, we observed O2 desorption in the LT range directly after
Conclusion and outlook
The main conclusions of this detailed STM and TPD study can be summarized as follows:
- i)
O2 molecules dissociate at Obr vacancy sites on the surface of bulk-reduced TiO2(110) crystals even at temperatures as low as 120 K (first channel). Because most of the Obr vacancies are healed upon high O2 exposure at 100 K ≤ T ≤ 150 K, the adsorbates, O2 molecules and Oot adatoms, are stabilized on the surface through withdrawal of electronic charge essentially from the near-surface region.
- ii)
By quantifying the
Acknowledgements
We gratefully acknowledge the support from the Danish Research Agency to the iNANO center, The Strategic Research Council, the Villum Kahn Rasmussen Foundation, The Carlsberg Foundation, and the European Research Council through an Advanced ERC grant. We are grateful to J. T. Yates Jr. and M. A. Henderson and for the stimulating discussions.
References (86)
- et al.
Chem. Soc. Rev.
(1993) - et al.
Chem. Rev.
(1993) - et al.
Prog. Solid State Chem.
(2004) Surf. Sci. Rep.
(2003)Bull. Chem. Soc. Jpn
(2004)- et al.
Chem. Soc. Rev.
(2008) - et al.
Surf. Sci. Rep.
(2008) - et al.
Chem. Rev.
(2006) - et al.
J. Chem. Phys.
(1995) - et al.
Langmuir
(1997)
Surf. Sci.
J. Phys. Chem. B
J. Phys. Chem. B
J. Phys. Chem. B
Chem. Phys. Lett.
Top. Catal.
J. Phys. Chem. B
J. Phys. Chem. B
J. Phys. Chem. C
Phys. Rev. Lett.
Phys. Rev. B
J. Phys. Chem. C
Phys. Rev. B
The Surface Science of Metal Oxides
Curr. Opin. Solid State Mater. Sci.
Prog. Surf. Sci.
J. Catal.
Surf. Sci.
Surf. Sci.
Surf. Sci.
Science
Phys. Rev. B
Phys. Rev. Lett.
Science
Surf. Sci.
J. Phys. Chem. B
ACS Nano
J. Phys. Chem. C
J. Phys. Chem. C
J. Phys. Chem. C
Phys. Chem. Chem. Phys.
J. Chem. Phys.
Cited by (131)
Photon Stimulated Desorption of C<inf>6</inf>H<inf>6</inf> on Rutile TiO<inf>2</inf>(110): Hole or Electron Regulated?
2024, Journal of Physical Chemistry C